[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 528.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: bitcoinvideos on May 05, 2014, 03:48:43 PM

v 0.7 from a few days ago compiles fine on Ubuntu...I've ran autogen and then compile then make with the same results.

okay, last lines in Makefile.am should say this instead.

Code:

# ABI requiring code modules
# this module doesn't compile with Compute 2.0 unfortunately
JHA/cuda_jha_compactionTest.o: JHA/cuda_jha_compactionTest.cu
$(NVCC) -I . -I cudpp-2.1/include @CFLAGS@ -Xptxas "-abi=yes -v" -gencode=arch=comp

it was a copy&paste from our internal SVN repo where we use a different filename, sorry.

bitcoinvideos

sr. member

Activity: 251

Merit: 250

v 0.7 from a few days ago compiles fine on Ubuntu...I've ran autogen and then compile then make with the same results.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: skunk on May 05, 2014, 03:27:23 PM

it fails to build on linux:

Code:

ptxas warning : 'option -abi=no' might get deprecated in future
ptxas info : 0 bytes gmem, 464 bytes cmem[3]
ptxas info : Compiling entry function '_Z26jackpot_keccak512_gpu_hashijPy' for 'sm_35'
ptxas info : Used 78 registers, 336 bytes cmem[0]
/opt/cuda/bin/nvcc -g -O2 -I . -Xptxas "-abi=no -v" -gencode=arch=compute_20,code=\"sm_20,compute_20\" -gencode=arch=compute_30,code=\"sm_30,compute_30\" -gencode=arch=compute_35,code=\"sm_35,compute_35\" --maxrregcount=80 --ptxas-options=-v -o JHA/cuda_jha_compactionTest.o -c JHA/cuda_jha_compactionTest.cu
JHA/cuda_jha_compactionTest.cu(117): error: identifier "__shfl_up" is undefined

JHA/cuda_jha_compactionTest.cu(144): error: identifier "__shfl_up" is undefined

2 errors detected in the compilation of "/tmp/tmpxft_00001dc7_00000000-13_cuda_jha_compactionTest.compute_20.cpp1.ii".
make[2]: *** [JHA/cuda_jha_compactionTest.o] Error 2
make[2]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make: *** [all] Error 2

cuda is 5.5.22 and before today's commits it was building fine...

can't build this module with compute 2.0, and also it needs abi=yes flag. I have it in Makefile.am like this.
Run autogen.sh maybe?

bigjme

sr. member

Activity: 350

Merit: 250

so it seems the psu i am using on my mining cards is dieing, the noise is horrible, time to order a new psu

djm34

legendary

Activity: 1400

Merit: 1050

interesting coin just launched: BlackDragon (scrypt-jane N=9) interesting reward mechanism (D&D like)

EtBIM

member

Activity: 83

Merit: 10

Quote from: THeZoiD on May 05, 2014, 03:31:27 PM

ok now doing 8.3mh/s with gtx770
Cheesy

Are you using x86 or x64 ? What are your clocks ? I can't get past 4.7 Mh/s with mine

Tyr808

sr. member

Activity: 607

Merit: 278

06/19/11 17:51 Bought BTC 259684.77 for 0.0101

Quote from: bigjme on May 05, 2014, 03:17:35 PM

Compute 3.5 - http://files.crimsonskulls.co.uk/ccminer%20-%2005-05-14%20-%20compute%203.5.zip
Compute 3.0 - http://files.crimsonskulls.co.uk/ccminer%20-%2005-05-14%20-%20compute%203.0.zip
Compute 2.0 - wont compile for some reason?

guys i dont know if i changed the groestl stuff properly so use on jackpot only please

Thanks for the compile! Still haven't had the time to tech myself how to do this yet.

My 750ti rig went from a maximum 2.9 mh/s to a massive 3.7 mh/s and is averaging at 3.5 mh/s. Much appreciation there.

Unfortunately, my 660ti rig losses hashes this build. I go from 3.9 mh/s on Christians last official binary (x86) to 3.4 mh/s on this new release. Double checked to make sure I used the right compute value too. Either way, NP complaints though, so cheers

EtBIM

member

Activity: 83

Merit: 10

I'm getting the same hashrate with compute 3.0, even a bit lower

liomojo1

hero member

Activity: 676

Merit: 500

Well this is the first time i see 64 bit version having the same sometimes better hash than x86. What i saw here is that droping the mem clock down makes the hash drop too. So if you want 4.1mhs you have too keep your mem clock at 3170 mhs and core clock 1380 . My max for now is 4180khs Grin

bigjme

sr. member

Activity: 350

Merit: 250

Quote from: skunk on May 05, 2014, 03:35:17 PM

i need compute 3.0, do you know if there is a way to build just for it under linux? i can't find an option for the configure script...

im not entirely sure sorry, i will have a think and see if i can find anything

skunk

sr. member

Activity: 329

Merit: 250

Quote from: bigjme on May 05, 2014, 03:27:44 PM

Quote from: skunk on May 05, 2014, 03:27:23 PM

cuda is 5.5.22 and before today's commits it was building fine...

it seems there is an issue with compute 2.0, and linux compiles all versions

i need compute 3.0, do you know if there is a way to build just for that under linux? i can't find an option for the configure script...

THeZoiD

sr. member

Activity: 476

Merit: 250

ok now doing 8.3mh/s with gtx770
Cheesy

bitcoinvideos

sr. member

Activity: 251

Merit: 250

Quote from: skunk on May 05, 2014, 03:27:23 PM

it fails to build on linux:

Code:

ptxas warning : 'option -abi=no' might get deprecated in future
ptxas info : 0 bytes gmem, 464 bytes cmem[3]
ptxas info : Compiling entry function '_Z26jackpot_keccak512_gpu_hashijPy' for 'sm_35'
ptxas info : Used 78 registers, 336 bytes cmem[0]
/opt/cuda/bin/nvcc -g -O2 -I . -Xptxas "-abi=no -v" -gencode=arch=compute_20,code=\"sm_20,compute_20\" -gencode=arch=compute_30,code=\"sm_30,compute_30\" -gencode=arch=compute_35,code=\"sm_35,compute_35\" --maxrregcount=80 --ptxas-options=-v -o JHA/cuda_jha_compactionTest.o -c JHA/cuda_jha_compactionTest.cu
JHA/cuda_jha_compactionTest.cu(117): error: identifier "__shfl_up" is undefined

JHA/cuda_jha_compactionTest.cu(144): error: identifier "__shfl_up" is undefined

2 errors detected in the compilation of "/tmp/tmpxft_00001dc7_00000000-13_cuda_jha_compactionTest.compute_20.cpp1.ii".
make[2]: *** [JHA/cuda_jha_compactionTest.o] Error 2
make[2]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make: *** [all] Error 2

cuda is 5.5.22 and before today's commits it was building fine...

I have the same compile error Ubuntu 64bit

bigjme

sr. member

Activity: 350

Merit: 250

Quote from: skunk on May 05, 2014, 03:27:23 PM

cuda is 5.5.22 and before today's commits it was building fine...

it seems there is an issue with compute 2.0, and linux compiles all versions

Quote from: sdesbure on May 05, 2014, 03:26:25 PM

well, I compiled myself and after used bigjme cudaminer but I'm stuck around 3MH/s :/ (I've got peak at 3.8Mh/s but the hash rate is moving fast from 2.1 to 3.8 MH/s)

are you mining on a pool? waited for diff to settle?

skunk

sr. member

Activity: 329

Merit: 250

it fails to build on linux:

Code:

ptxas warning : 'option -abi=no' might get deprecated in future
ptxas info : 0 bytes gmem, 464 bytes cmem[3]
ptxas info : Compiling entry function '_Z26jackpot_keccak512_gpu_hashijPy' for 'sm_35'
ptxas info : Used 78 registers, 336 bytes cmem[0]
/opt/cuda/bin/nvcc -g -O2 -I . -Xptxas "-abi=no -v" -gencode=arch=compute_20,code=\"sm_20,compute_20\" -gencode=arch=compute_30,code=\"sm_30,compute_30\" -gencode=arch=compute_35,code=\"sm_35,compute_35\" --maxrregcount=80 --ptxas-options=-v -o JHA/cuda_jha_compactionTest.o -c JHA/cuda_jha_compactionTest.cu
JHA/cuda_jha_compactionTest.cu(117): error: identifier "__shfl_up" is undefined

JHA/cuda_jha_compactionTest.cu(144): error: identifier "__shfl_up" is undefined

2 errors detected in the compilation of "/tmp/tmpxft_00001dc7_00000000-13_cuda_jha_compactionTest.compute_20.cpp1.ii".
make[2]: *** [JHA/cuda_jha_compactionTest.o] Error 2
make[2]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/skunk/bitcoin/ccminer2'
make: *** [all] Error 2

cuda is 5.5.22 and before today's commits it was building fine...

sdesbure

newbie

Activity: 6

Merit: 0

Quote from: Breadpudding on May 05, 2014, 03:14:53 PM

Confirming that my 750ti is getting about a 1MH/s increase (to 4.1MHs) also on Jackpot coin solo mining & pool mining, for both x86 and x64 version.

Great job C&C hashfactory!

Quote from: bigjme on May 05, 2014, 02:41:42 PM

fired te compile up with my 780 while others are busy and shares being accepted. my 780 is mining at over 7MH/s

seems to be exactly a 1MH/s increase

(

well, I compiled myself and after used bigjme cudaminer but I'm stuck around 3MH/s :/ (I've got peak at 3.8Mh/s but the hash rate is moving fast from 2.1 to 3.8 MH/s)

bigjme

sr. member

Activity: 350

Merit: 250

Quote from: Breadpudding on May 05, 2014, 03:23:44 PM

bigjme, just sent a 40k JPC donation your way for all the compiling work that you do Cheesy

BP

much appreciated

Breadpudding

member

Activity: 76

Merit: 12

bigjme, just sent a 40k JPC donation your way for all the compiling work that you do Cheesy

BP

bigjme

sr. member

Activity: 350

Merit: 250

Compute 3.5 - http://files.crimsonskulls.co.uk/ccminer%20-%2005-05-14%20-%20compute%203.5.zip
Compute 3.0 - http://files.crimsonskulls.co.uk/ccminer%20-%2005-05-14%20-%20compute%203.0.zip
Compute 2.0 - wont compile for some reason?

guys i dont know if i changed the groestl stuff properly so use on jackpot only please

antonio8

legendary

Activity: 1400

Merit: 1000

Quote from: Chris84 on May 05, 2014, 01:51:31 PM

Quote from: antonio8 on May 05, 2014, 12:17:05 PM

Quote from: THeZoiD on May 05, 2014, 12:06:54 PM

Quote from: antonio8 on May 05, 2014, 12:01:51 PM

@ C&C or anyone else

Question about CCMiner or issue I am having.

Does anything else special need to be installed installed on the computer to run properly?

On my rig 1 I have 3 750ti's and 1 670 and everything runs smooth as butter.

On rig 2 I have 5 750ti's and I can not get over 500 kh/s on any card whether I run them individual, all together or in groups.

I basically copied the folder from rig 1 to rig 2 and made .bat files. I just changed the -d on rig 2 in the bat files to correspond to the gpu's. Here is my example:
ccminer50.exe -a mjollnir -d 0
ccminer50.exe -a mjollnir -d 0,1,2

my question is how does it run smooth when gtx670 with ccminer50?
ccminer50 is for maxwell gpu

That was for the 750 ti's. The 670 runs on ccminer30.exe

If you are sure, that the devices specified with -d are correct and the 670 IS NOT INCLUDED, then plz check your cpu-usage. Is this the limiting factor?
On rig you have 5x 750Ti and no other Cards? then specify: "ccminer50.exe -a mjollnir -t 5 -d 0,1,2,3,4" (maybe you don't Need -d Option, if you're using every Card in the System, but I'm not sure, didn't test it).

Christian

Thanks for the reply.

Will try this in about 10 hrs when I get home.

I did not put the -t in the bat but will try it. I will try just -t 5 and the -t and -d and see how it goes. I also did not pay attention to the cpu usage as I know it is already at 100% solo mining a "cpu only" coin. I'll turn that off also and recheck.

This is for my rig with only 5 of the 750ti's.

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 528. (Read 3426976 times)